Time series analyses with psychometric data

Area of Science:

• Psychology
• Psychosomatic Medicine
• Quantitative Research Methods

Background:

• Understanding interactional dynamics is crucial in psychology and psychosomatic medicine.
• Longitudinal studies in these fields often face challenges with missing values and non-equidistant measurements.
• Existing process-oriented methodologies are typically designed for equidistant data on ratio scales.

Purpose of the Study:

• To clarify the conditions for satisfactory performance of longitudinal methods with psychological and psychosomatic data.
• To examine the robustness of the Johansen test for time series analysis under various data quality conditions.

Main Methods:

• Utilized a Monte Carlo simulation method to assess the performance of the Johansen test.
• Evaluated the impact of sample size, data discreteness, measurement equidistance, and scale properties on time series analysis.

Main Results:

• Time series analyses require a minimum of 70 observations for accurate estimation and inference.
• Discrete data and non-equidistant measurements due to missing values generally do not pose significant problems.
• Stationary processes can be adequately captured with 5- or 7-point ordinal scales.
• Trending processes necessitate at least 10-point scales for acceptable estimation and inference quality.

Conclusions:

• The Johansen test demonstrates robustness under common data quality issues in psychological research.
• Recommendations are provided for sample size and scale selection to ensure reliable longitudinal data analysis.
• Findings offer guidance for researchers using time series methods with ordinal or interval data in psychology and psychosomatic medicine.